Regulating Artificial Intelligence in Finance: Lessons from Singapore (OpenAlex (recherche mondiale))

Summary

• According to KPMG, 71% of surveyed financial firms have deployed AI tools, with global banking sector AI spending estimated at USD 31.3 billion in 2024.p.6
• DBS deployed more than 1,500 AI models across over 370 use cases, generating over SGD 750 million in economic value in 2024.p.6
• GXS Bank's eBuddy chatbot reduced call volumes by approximately 20%, while DBS projects its CSO Assistant will shorten call durations by up to 20%.p.12
• DBS detected over 95% of non-performing SME loans at least three months before credit stress occurred in 2022, helping over 80% of at-risk borrowers avoid default.p.14
• In a 2024 Cambridge SupTech Lab study, 80% of 139 surveyed authorities reported that SupTech was effective in improving data-driven decision-making.p.38

Key insights

1. Singapore governs AI in finance through a regulator-led, collaborative strategy using soft law instruments established by the Monetary Authority of Singapore.p.7
2. AI helps resolve the Financial Inclusion Trilemma by reducing operational costs, enabling precise customer profiling, and enhancing fraud detection.p.9
3. UOB's AI solution operates with a 96% true positive rate while processing over 5,700 transaction alerts each month for anti-money laundering.p.11
4. Transparency challenges are heightened when financial institutions use third-party AI models whose proprietary details are withheld to protect intellectual property.p.12
5. GenAI hallucinations create severe risks when deployed in mission-critical customer-facing financial services without human oversight.p.15
6. The non-binding nature of the FEAT Principles presents a structural limitation that can lead to inconsistent application across financial institutions.p.21
7. MAS issued a consultation paper on proposed Guidelines on AI Risk Management in November 2025 to set clearer operational standards for financial institutions.p.24
8. Deploying supervisory technology allows regulators to automate routine data collection and improve oversight of cybersecurity and credit risks.p.38
9. A shared responsibility framework addresses third-party AI risks by allocating pre-deployment compliance to developers and post-deployment compliance to financial institutions.p.40